You are on page 1of 19

FACE RECOGNITION USING LDA WITH WAVELET TRANSFORM APPROACH

BY B PRASHANTH Click to edit Master subtitle style 20080785 S NAVEEN 20080776 S VENUGOPAL SUGANTHAN Under 20080817 the guidance of Mrs. A Viji Saravanan Dept. of Electronics
4/27/12

OBJECTIVE

The objective of this project is to perform face recognition using the algorithms such as Principal component analysis(PCA), Linear Discriminant Analysis(LDA) and its variants. The above results are to be compared using LDA with wavelet transform approach and Independent component Analysis(ICA).

4/27/12

START HAAR WAVELET TRANSFORM CALCULATE GLOBAL MEAN CALCULATE S b AND S w COMPUTE THE WEIGHT OF THE IMAGE
COMPARE THE WEIGHTS OFTEST IMAGE AND TRAINNG IMAGE

STOP

4/27/12

WORK DONE:

Finished the codes of component analysis, discriminant analysis. Obtained the graphs with ATT and IFD in the x axis accuracy, total time, training testing time in the y axis.

Principal Linear Datasets and the time and

WORK TO BE DONE:

Proposed algorithm of Linear discriminant analysis with Haar Wavelet transform. 4/27/12 Comparison of all the results.

All the manipulations for this project are done using the below hardware configurations and software tools Processor Clock speed RAM Matlab version

Intel core 2 duo 2.00 Ghz 3 Gb R2008

4/27/12

ATT DATABASE:

IFD DATABASE:

4/27/12

Principal Component Analysis


This technique is used to approximate the original data with lower dimensional feature vectors. The basic approach is to compute the eigenvectors of the covariance matrix and approximate the original data by the linear combination of leading eigenvectors. The covariance matrix is calculated by the C= given formula,
of where denotes the average 4/27/12

PCA OUTPUT.

IP IMAGE

AFTER SUBTRACTING THE MEAN

EIGEN VECTORS

4/27/12

4/27/12

4/27/12

4/27/12

4/27/12

Linear Discriminant Analysis


It performs dimensionality reduction while preserving as much of the class discriminatory information as possible. It seeks to find directions along which the classes are best separated. It takes into consideration the scatter withinclasses but also the scatter between-classes and is more capable of distinguishing image variation due to identity Sb = from variation due to other sources such as illumination and expression.
Sw = 4/27/12

4/27/12

4/27/12

4/27/12

4/27/12

References
1) Neeta Nain, Akshay Kumar, Amlesh Kumar Mohapatra, Ratan Das, Ashok Kumar& NemiChand Singh , FACE RECOGNITION USING LDA WITH WAVELET TRANSFORM APPROACH International journal on information technology and knowledge management, July-December 2011, Volume 4, pp. 603-607. 2) M.A.Turk and A.P.Pentland , Face Recognition Using Eigen faces, IN Proc.of Computer Vision and Pattern Recognition,pages 586-591,IEEE, June 1991b. 1) M.Loog, R.P.W.Duin, R.Hacb-Umbach, Multiclass Linear Dimension Reduction 4/27/12 by Weighted Pairwise

THANK YOU
Click to edit Master subtitle style

4/27/12

You might also like